Decision criteria for when to stop forward testing: sufficient sample size, consistent metrics, and statistical confidence thresholds.